To Sub or not to Sub the RF-7, that is the question.

[mas]

Well you're never going to get it figured out trying to use any form of EQ...

The problem is that the bass issues in your room exist in the time-domain (reflections and all of the standing waves and all of the superposition happen because the sound takes time to bounce around the room). An EQ is a frequency-domain device and does not correct time-domain issues.

If you're interested, I can point you towards some time-domain solutions. If not, well I hope you enjoy beating your head against the wall []

Nonsense! I can EQ my way out of any problem!

In fact, this weekend I even used it to solve a problem with my grass. ...Now to go and use it to keep my cat from licking me on the nose while I am sleeping!

But what you do not want to do is to check out the Interfacing a subwoofer with a full range system - or the Active Crossover article in the Technical section.

Beyond that, you want to address the room modes...

[dbossa]

Getting the bass right is the hardest freakin' thing I think I've ever had to figure out in my life.... seriously.

Well you're never going to get it figured out trying to use any form of EQ...

The problem is that the bass issues in your room exist in the time-domain (reflections and all of the standing waves and all of the superposition happen because the sound takes time to bounce around the room). An EQ is a frequency-domain device and does not correct time-domain issues.

If you're interested, I can point you towards some time-domain solutions. If not, well I hope you enjoy beating your head against the wall []

Yes, please do, by all means. I was just trying to get the best end result with what I already had vs spending more money [:$]

[DrWho]

Getting the best results with what we have is what it's all about. [Y]

I hope I didn't offend...I was just reminiscing on my past experiences

with EQ (and recent experiences undoing the damage).

The

absolute best thing you could do for your system is to employ various

forms of acoustical treatment. Since we're talking about the low

frequencies, you want to look at bass traps and helmholtz resonators.

The treatments don't really get that expensive until you start trying

to make everything look pretty. Everyone's situation is going to be

different in that regard so it becomes difficult to recommend specific

products (not to mention you'll want to base them off the measurements

anyway). Nevertheless, it's the only way you can achieve consistent

behavior throughout the room.

As far as your fancy EQ box and

your measurements with the SPL meter...an understanding of the time

domain and measurement process should aid your decisions about the

filters. This would be a lot easier if you could provide the

measurements you've been obtaining with the program...uploading the

impulse response would be ideal since that's what every measurement is

derived from. Anyways, the following is an approach I've been

experimenting with and actually having pretty favorable results to:

You

never want to try and boost at frequencies where you have nulls because

the laws of superposition dictate that you can never "fill the void"

with EQ. Yes, you can play with EQ and make the measurement indicate

that the void is filled, but all you've done is dramatically increased

the amplitude of that frequency before the reflection arrives and

cancels itself out. Since the frequency response calculation is taking

a blind sum of all the values inside the specified window, it doesn't

reveal this behavior in the frequency response (man I say that

loosely). What you end up doing is boosting the direct sound far beyond

a natural level, which when combined with the near sudden decay due to

the acoustical interference will sound even more unnatural.

This

can be understood a bit better when you consider the "Haas window".

It's basically a measure of the longest amount of time between two

sounds for them to be percieved as two distinct sounds - so if you have

a reflection within the Haas window, then you hear the direct and

reflected sound as a single sound. If the reflection happens outside

the Haas window (say > 20-40ms), then it gets percieved as an echo

seperate from the direct sound. Since our low frequency measurement

spans a long period of time, we are seeing the combined total of the

direct and reflected sound. However, I would argue that our ears have

the ability to compensate and effectively ignore the acoustics of the

room....but only those acoustics that exist outside the Haas window. In

other words, I would argue that EQ doesn't need to (and ultimately

can't) compensate for aberrations in the response that occur outside

the Haas window (except any behavior that will sound especially

unnatural....like super fast or very slow decay rates, or distinct

reflections). In other words, I would argue that any amount of EQ that

we do to the low frequencies needs to only account for what occurs

within the Haas window.

I know I'm really butchering a lot of

concepts to get to this point, but assuming everything I've said above

is along the right track, the situation is still very hopeless. In

order to achieve a high resolution measurement at the lower

frequencies, you need to have an increasingly longer window on the

impulse response...say on the order of 500+ ms. This is way outside

even a 50ms Haas window which means there is no way to accurately

distinguish the behavior that our ears are able to filter out. Try

shortening the impulse response window and you'll see the frequency

response gradually becoming more smoothed out (less resolution) at the

low-end. This is an issue because you want the Q of the filter to match

the Q of the problem in your room (in other words, you only want to

attenuate the problem frequencies....not neighboring frequencies that

don't need fixing). In order to determine the Q of the room problem,

you must use a long window. However, doing this is going to

dramatically change the amplitude you see on the graph. To get around

this, I've been experimenting with using a long window to determine the

Q, and then using a short filter to get an idea of the amplitude of the

problem. Some people just try to attenuate by half of the total

problem, but I think you can achieve more meaningful data by trying to

shorten the window to the range of the Haas window to get an idea of

what kinds of amplitudes you're dealing with. So pick the Q with a long

window, and the magnitude of attenuation with the short window and you

should end up with something that sounds better than it did

before...ideal? certainly not, but EQ already isn't ideal anyway...

It should be noted that this procedure will never solve

the acoustical problem in the room, but it should reduce in magnitude

the negative perceptions of those sounds. It's kinda like a band-aid.

It should also be noted that it's only going to 'work' at one single

listening position. I'm sure you've already noticed how moving the mic

a few inches can swing the frequency response +-10dB []

I

really hesitate suggesting EQ methods because in principal it's an

invalid approach and $100 of acoustical treatment is going to have a

more beneficial impact. Nevertheless, I can't ignore that there is some

benefit to using an EQ and that different EQ is employed with the same

speakers in different acoustical environments. The above is just my

rational for quantifying the magnitude of good effect an EQ can always

have on a system...now I'm just waiting for an instance where it

doesn't improve and then quantifying the broken concept to account for

it....or you could just solve the problem the correct way in the first

place with acoustical treatment when the EQ thing doesn't work out []

Anyways, I hope this helps and I welcome any criticisms and reports of people trying it out.

[Islander]

Okay, I can follow that you can't EQ up to "fill a void". What about EQing down to quiet or reduce an unwanted peak? If less energy (at the frequency of the resonance) is coming into the room and the room gain flattens a dip caused by the EQ, is that logical?

[dbossa]

Okay, I can follow that you can't EQ up to "fill a void". What about EQing down to quiet or reduce an unwanted peak? If less energy (at the frequency of the resonance) is coming into the room and the room gain flattens a dip caused by the EQ, is that logical?

I believe that, according to what he said, that would be fine. It's boosting the nulls that can make the sound come out sounding a touch unnatural. I think that could be one of the culprits behind my problems as I boosted a few nulls.

I also noticed that the volume affects my graph. Setting the volume to 75db and then setting the volume to 85db gives completely different graphs. Seeing as how I tend to listen to louder levels when using my sub, I wonder if I should be taking measurements at louder levels since they are more realistic vs the suggested 75db?

[DrWho]

Is the peak the result of resonance in the time-domain, or the result of a spike in the response without resonance?

If it's a resonance, then you've got the same Haas window issue. If you attenuate the resonant frequency so that it measures flat over say 500ms, then it's not going to measure flat for the first 40ms. In other words, the frequency response of the system changes over time as new reflections arrive. Let's say the output of your speakers is perfectly flat, then in the room it measures +3dB at 45Hz over the first 30ms, and then +10dB at over 500ms. If you attenuate -10dB at 45Hz, your actual direct sound is going to be 10dB too quiet and the perceived direct sound is going to be 7dB too quiet. The resonance may not be as noticeable or as annoying, but you totally destroy the direct sound in the process - something our ears won't be able to ignore as readily.

Instead, let's say you attenuate only -3dB at 45Hz. The actual direct sound of the speakers will be compromised, but the perceived direct sound will be flat. You'll measure the output being +6dB over 500ms, but you will perceive it as room related and should, in theory, be able to ignore the detrimental effects of the room. It will boom less than unattenuated, but more than if you attenuated it more.

It's kinda like hitting a bell - hit it hard and it rings forever. Hit

it slightly softer and it doesn't seem to ring as much, but also sounds quieter initially. The ring actually

decays at the same rate, but just gets washed up in the noise floor

sooner. If you hit it so soft that it doesn't ring, then you're not

going to hear the bell at all. By gating the impulse

response to look at what happens over the Haas Window, you can see

how much attenuation the perceived direct sound needs...basically making it so that the initial strike sounds like its at the same volume.

I guess you could just break the sound up into 3 segments: You have the original output from the speakers, the combination of direct and reflected sound over the Haas Window, and then all the sound that comes after the Haas Window ("reverberant field"). I keep referring to the Haas Window and not a specific time because it seems that the actual time periods involved are frequency and source material dependent. For example, it's going to be longer for smooth legato low pitched singing recorded in a cathedral than it will be for short and snappy mids, like pizzicato recorded in iso-booths. The point isn't so much as to concentrate on numbers as it is to understand the concept and how each section of the time domain is percieved...

So where you've got destructive interference, any attenuation is going to make the direct sound too loud and then result in a very fast decay. It won't sound like your room is being filled with sound, the sub will be working much much harder, and distortion will increase. Where you've got resonance, any attenuation is going to make the direct sound lose visceral impact and sound weak on transients, yet still booming a bit in the room...it basically sounds sloppy. At least that's my take on it...

Sorry for rambling so much.

[DrWho]

I also noticed that the volume affects my graph. Setting the volume to 75db and then setting the volume to 85db gives completely different graphs. Seeing as how I tend to listen to louder levels when using my sub, I wonder if I should be taking measurements at louder levels since they are more realistic vs the suggested 75db?

It's like the bell...when you listen louder, you excite the resonances of the room harder, and so they ring longer. Since the windows on your measurements default to around 500ms(?), you've got more of the tail end of the resonance making the response look louder on the measurement. If you gated your response at 75dB and 85dB to show the percieved direct sound and then attenuated accordingly (say a gated response around 50ms), then the balance should be percieved as more consistent at the various listening levels.

[dtel]

Mas................. " In fact, this weekend I even used it to solve a problem with my grass."

Was it the " EQ " or the " Grass " that led you to think you solved a problem ? []

[zuzu]

Before I realized my new velodyne DD 12 was too small for my 4000cu space and too inefficient for my RF 7's base reflex design, I played with the on screen software. Each day I would look at my best EQ settings from the day before and discover they were not nearly the best for today. Especially if I moved the mike a few inches up down sideways. Other things like sub volume level, low crossover point and slope and phase/polarity were much more consistent from day to day. DR Who? What is Q? What is Contour. Never got that far in playing with the settings with the software.

[dbossa]

OK, I have much better results now even though I haven't added any room treatments (which I'll do eventually). This is what I did:

REW gives a target of 75db. I noticed that performing sweeps using a target of 75db and then doing more at 85db gave completely different results. Since I tend to listen to music and movies at louder levels I used a target of 85db instead. Using the sweep is not effective in a room that has poor acoustics for bass. What I did I took manual readings using REW and let each frequency sit for a good two seconds to allow the bass to reverberate through the room which would usually cause certain frequencies to rise a lot higher vs doing the sweep. Once I had my graph I made my adjustments and made sure that my bass would roll off after a certain frequency just like the curve that REW displays. So, I essentially followed the same curve only 10db's above it. Also, I did not add any gains of any kind to increase the nulls. I simply got rid of the peaks. The bass is now far more balanced though a touch lacking in some areas. However, I find that it sounds much better than having too much bass reverberating through the room. In some places where I used to reduce a peak by 5db when using the sweep, I had to reduce the same frequency by 15db because I allowed it to resonate throughout the room. This became a good band-aid until I get some bass traps.

[Islander]

Is the peak the result of resonance in the time-domain, or the result of a spike in the response without resonance?

Instead, let's say you attenuate only -3dB at 45Hz. The actual direct sound of the speakers will be compromised, but the perceived direct sound will be flat. You'll measure the output being +6dB over 500ms, but you will perceive it as room related and should, in theory, be able to ignore the detrimental effects of the room. It will boom less than unattenuated, but more than if you attenuated it more.

So where you've got destructive interference, any attenuation is going to make the direct sound too loud and then result in a very fast decay. It won't sound like your room is being filled with sound, the sub will be working much much harder, and distortion will increase. Where you've got resonance, any attenuation is going to make the direct sound lose visceral impact and sound weak on transients, yet still booming a bit in the room...it basically sounds sloppy. At least that's my take on it...

Did you mean "attenuation" or "augmentation" in the sentence referring to destructive interference?

Mike, I quoted only the parts that bear directly on my listening situation. A few weeks ago, I decided to connect my old unused equalizer to my sub to try to tame a resonance in the room. Lowering the response 5dB at 30Hz did exactly as you just described. The sound was less annoying, but a little weak and thin-sounding, showing that the measurements I'm using don't tell the whole story.

However, I've got a new sofa and chair arriving soon, with taller backs, that may have an effect on the room response, so after they're set up (if they need to break in, I'm thinking it'll take place over months or years []), I'll turn off the sub and start from scratch.

It sounds like the best compromise, after the room is treated as best I can manage and the sub is in its best location with its best settings, is to only partially try to correct room resonances with the EQ, since what I hear is more important than what I measure. Is that roughly correct?

BTW, thanks to you and mas for explaining audio theory. Some of it is is actually starting to sound like English to me.

[Buckeye_Nut]

Note: my room is 1700 sq ft (dorm room),

Jeezus.... You must have a dorm room like Thornton Melon's in the movie "Back to School".

Ya got a 10 person jacuzzi in there too?

[H]

[DrWho]

Did you mean "attenuation" or "augmentation" in the sentence referring to destructive interference?

Attenuation as in what you would achieve with an EQ. Boosting that particular frequency with an EQ is going to make the direct sound too loud.

A few weeks ago, I decided to connect my old unused equalizer to my sub to try to tame a resonance in the room. Lowering the response 5dB at 30Hz did exactly as you just described. The sound was less annoying, but a little weak and thin-sounding, showing that the measurements I'm using don't tell the whole story.

Ya, you really need to obtain a waterfall response of your room and then it'll become very clear. I should post some graphs when I get the chance...

It sounds like the best compromise, after the room is treated as best I can manage and the sub is in its best location with its best settings, is to only partially try to correct room resonances with the EQ, since what I hear is more important than what I measure. Is that roughly correct?

I would argue that measurements always correlate to what you hear. It's how one interprets the measurements that matters...basically understanding the limitations of the measurement technique. It would be silly to measure the color of a tree with an osciliscope. It's just as obnoxious to measure room acoustics with an RTA or any other non-time-based technique (like test tones and an SPL meter). In other words, we need to train our ears.

BTW, thanks to you and mas for explaining audio theory. Some of it is is actually starting to sound like English to me.

Great! Perhaps you can start translating for me then []

[Jay481985]

Note: my room is 1700 sq ft (dorm room),

Jeezus.... You must have a dorm room like Thornton Melon's in the movie "Back to School".

Ya got a 10 person jacuzzi in there too?

[H]

That's huge, I have a pair of rf-83 and a rsw12 in a 150 sq foot room.

[masterxela]

?!?!, my dorm room is 300 sq ft. Either he has a 6 person job, or thats cubic feet.....

[Arky]

Yeah, that's the size of a house.